Content and creative are where most marketing teams start with AI, because the work is high-volume and the savings are easy to see.
Where AI helps in the content chain
- Research and briefs: summarising customer insight, competitor content and past results into a brief.
- First drafts and variants: headlines, body copy, email versions, social posts, localisation.
- Images and video: product shots, backgrounds, seasonal versions, resizing for every channel, short cut-downs.
- Review support: checking copy against brand rules, claims and reading level before a human reviews it.
What results look like when it works
Klarna said in May 2024 that generative AI saved it about $10 million a year in marketing: around $6 million in image production (using tools such as Midjourney, DALL-E and Firefly), with the image development cycle cut from six weeks to seven days, and a 25% cut in spending on external marketing suppliers. It reported making more than 1,000 AI images in the first quarter of 2024 and using an in-house tool for most copywriting. These are company-reported figures.
Unilever says it uses digital twins of its products (built with NVIDIA Omniverse) to produce product imagery two times faster and 50% cheaper, with consistent packs in every image.
What to watch
- Quality control becomes the bottleneck: more content means more to review. Build checks into the workflow.
- Brand drift: without a voice guide and examples, AI copy sounds like everyone else's.
- Rights: use tools and inputs you have rights to; keep records.
- Savings leak: if nobody decides what to do with freed time and money, it disappears into more volume.
Start here
Pick one high-volume asset type (social variants, product images, email versions), measure cost and cycle time for a month, introduce an approved AI workflow with human review, and compare.